Is Deepseek Safe?
Deepseek — Nerq Trust Score 37.2/100 (F grade). Based on analysis of 2 trust dimensions, it is has significant safety risks. Last updated: 2026-03-31.
Exercise caution with Deepseek. Deepseek is an AI tool with a Nerq Trust Score of 37.2/100 (F), based on 3 independent data dimensions. It is below the recommended threshold of 70. Security: 65/100. Popularity: 5/100. Data sourced from privacy policy analysis, data handling practices, company background, and security certifications. Last updated: 2026-03-31. Machine-readable data (JSON).
Is Deepseek safe?
NO — USE WITH CAUTION — Deepseek has a Nerq Trust Score of 37.2/100 (F). It has below-average trust signals with significant gaps in security, maintenance, or documentation. Not recommended for production use without thorough manual review and additional security measures.
What is Deepseek's trust score?
Deepseek has a Nerq Trust Score of 37.2/100, earning a F grade. This score is based on 2 independently measured dimensions including security, maintenance, and community adoption.
What are the key security findings for Deepseek?
Deepseek's strongest signal is security at 65/100. No known vulnerabilities have been detected. It has not yet reached the Nerq Verified threshold of 70+.
What is Deepseek and who maintains it?
| Author | Unknown |
| Category | ai_tool |
| Source | N/A |
Popular Alternatives in ai_tool
Similar Ai Tool by Trust Score
Safety Guide: Deepseek
What is Deepseek?
Deepseek is a software — Open-source AI models by Chinese AI lab. DeepSeek-V2 competitive with top models at lower cost..
How to Verify Safety
Review the project for recent activity and known issues.
You can also check the trust score via API: GET /v1/preflight?target=DeepSeek
Key Safety Concerns for softwares
When evaluating any software, watch for: maintenance status, security.
Trust Assessment
Deepseek has a Nerq Trust Score of 37/100 (F) and has not yet reached Nerq trust threshold (70+). This score is based on automated analysis of security, maintenance, community, and quality signals.
Alternatives
- haotian-liu/LLaVA — 71/100
- wan22_i2v_14b_orbit_shot_lora — 59/100
- ChuckNorris (L1B3RT4S Prompt Enhancer) — 46/100
Key Takeaways
- Deepseek has a Trust Score of 37/100 (F).
- Review carefully before use — below trust threshold.
- Always verify independently using the Nerq API.
Detailed Score Analysis
| Dimension | Score |
|---|---|
| Security | 65/100 |
| Privacy | 36/100 |
| Reliability | 50/100 |
| Transparency | 40/100 |
| Maintenance | 60/100 |
Based on 5 dimensions. Data from privacy policy analysis, data handling practices, company background, and security certifications.
What data does Deepseek collect?
Deepseek is an AI tool. Open-source AI models by Chinese AI lab. DeepSeek-V2 competitive with top models at lower cost.
Privacy score: 36/100. AI tools may use inputs for model improvement unless explicitly opted out. Check the data usage policy before sharing confidential information, code, or personal data.
Consider whether the tool offers enterprise plans with data isolation, SOC 2 compliance, or on-premise deployment options.
Full analysis: Deepseek Privacy Report · Privacy review
Is Deepseek secure?
Security score: 65/100. Open-source AI models by Chinese AI lab. DeepSeek-V2 competitive with top models at lower cost.
Check Deepseek's security page for certifications such as SOC 2 Type II, ISO 27001, or GDPR compliance documentation. These certifications indicate that the vendor follows established security practices and undergoes regular audits.
For enterprise deployments, verify SSO/SAML support, role-based access control, and audit logging capabilities.
Full analysis: Deepseek Security Report
How we calculated this score
Deepseek's trust score of 37.2/100 (F) is computed from privacy policy analysis, data handling practices, company background, and security certifications. The score reflects 5 independent dimensions: security (65/100), privacy (36/100), reliability (50/100), transparency (40/100), maintenance (60/100). Each dimension is weighted equally to produce the composite trust score.
Nerq analyzes over 7.5 million entities across 26 registries using the same methodology, enabling direct cross-entity comparison. Scores are updated continuously as new data becomes available.
This page was last reviewed on March 31, 2026. Data version: 1.0.
Full methodology documentation · Machine-readable data (JSON API)
Frequently Asked Questions
Is Deepseek safe to use?
What is Deepseek's trust score?
What are the best alternatives to Deepseek?
Does Deepseek use my data for training?
Is Deepseek safe for confidential work?
Popular in ai_tool
Browse Categories
Disclaimer: Nerq trust scores are automated assessments based on publicly available signals. They are not endorsements or guarantees. Always conduct your own due diligence.